Corrupting Noise Estimation Based on Rapid Adaptation and Recursive Smoothing

  • François Xavier Nsabimana
  • Vignesh Subbaraman
  • Udo Zölzer
Conference paper
Part of the Communications in Computer and Information Science book series (CCIS, volume 130)


This work describes an algorithm that estimates the corrupting noise power from the speech signal degraded by stationary or highly non-stationary noise sources for the speech enhancement. The proposed technique combines the advantages of minimum statistics and rapid adaptation techniques to address especially low SNRs speech signals. In the first step, the algorithm starts the noise power estimation using minimum statistics principles with a very short adaption window. This yields an overestimation of the noise power that is finally accounted for using recursive averaging techniques. To ensure minimum speech power leakage into estimated noise power the algorithm updates the noise power using an unbiased estimate of the noise power from the minimum statistics approach. To outline the performances of the proposed technique objective and subjective grading tests were conducted for various noise sources at different SNRs.


Noise estimation Minimum statistics Recursive smoothing Rapid adaptation Voice activity Detection Speech presence probability Normalized mean square error 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Loizou, P.C.: Speech Enhancement Theory and Practice, 1st edn. Taylor and Francis Group, New York (2007)Google Scholar
  2. 2.
    Martin, R.: Spectral subtraction based on minimum statistics. In: Proc. of EUSIPCO (1994)Google Scholar
  3. 3.
    Martin, R.: Noise power spectral density estimation based on optimal smoothing and minimum statistics. IEEE Trans. on Speech, Audio Processing 9 (July 2001)Google Scholar
  4. 4.
    Cohen, I., Berdugo, B.: Noise estimation by minima controlled recursive averaging for robust speech enhancement. IEEE Signal Proc. Letters (January 2002)Google Scholar
  5. 5.
    Rangachari, S., Loizou, P.C.: A noise estimation algorithm for highly non stationary environments. In: Proc. of Speech Communications, vol. 48 (February 2006)Google Scholar
  6. 6.
    Cohen, I.: Noise spectrum estimation in adverse environments: improved minima controlled recursive averaging. IEEE Trans. Speech Audio Processing 411 (2003)Google Scholar
  7. 7.
    Rangachari, S., Loizou, P.C., Hu, Y.: A noise estimation algorithm with rapid adaptation for highly nonstationary environments. In: Proc. of ICASSP, vol. 1 (May 2004)Google Scholar
  8. 8.
    Erkelens, J.S., Heusdens, R.: Fast noise tracking based on recursive smoothing of mmse noise power estimates. In: Proc. of ICASSP (February 2008)Google Scholar
  9. 9.
    Erkelens, J.S., Heusdens, R.: Tracking of nonstationary noise based on data-driven recursive noise power estimation. IEEE Trans. on Audio, Speech, and Language Processing 16, 1112–1123 (2008)CrossRefGoogle Scholar
  10. 10.
    Doblinger, G.: Computationally efficient speech enhancement by spectral minima tracking in subbands. In: Proc. of Eurospeech, vol. 2 (September 1995)Google Scholar
  11. 11.
    Hu, Y., Loizou, P.C.: Evaluation of objective quality measures for speech enhancement, vol. 16, pp. 229–238 (January 2008)Google Scholar
  12. 12.
    Nsabimana, F.X., Subbaraman, V., Zölzer, U.: Noise power estimation using rapid adaptation and recursive smoothing principles. In: Proc. of the International Conference on Signal Processing and Multimedia Applications (SIGMAP 2009), Milan, Italy, July 7-10, pp. 13–18 (2009)Google Scholar
  13. 13.
    Nsabimana, F.X., Subbaraman, V., Zölzer, U.: A single channel speech enhancement technique exploiting human auditory masking properties. In: Journal of the 12th Conference for the International Union of Radio Science (URSI 2009), Miltenberg, Germany, September 28-October 01 (2009)Google Scholar
  14. 14.
    Nsabimana, F.X., Subbaraman, V., Zölzer, U.: A single channel speech enhancement technique using psychoacoustic principles. In: Proc. of the 17th European Signal Processing Conference (EUSIPCO 2009), Glasgow, Scotland, August 24-28, pp. 170–174 (2009)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • François Xavier Nsabimana
    • 1
  • Vignesh Subbaraman
    • 1
  • Udo Zölzer
    • 1
  1. 1.Department of Signal Processing and CommunicationsHelmut Schmidt UniversityHamburgGermany

Personalised recommendations